NU Smart Farms

Detecting lameness in commercial pig production

Background

The problem of lameness in pig production

Lameness in livestock is an issue for both animal welfare and sustainable food production. A lame animal may be in pain and reluctant to bear weight on the affected limb, meaning that it suffers from impaired mobility. Growth may be reduced and welfare affected, due to lower feed and water intake.

Lameness has many different causes. Some are genetic, some are environmental. Even well-managed farms can have some lameness in their pigs. Typically, up to 5% of pigs on a farm could be lame. But there are more subtle changes in walking patterns in up to 20% of animals. This points to underlying problems that cause changes in walking patterns and which may, in turn, lead to lameness.

We use motion capture technology from the film industry to develop automated ways of detecting lameness in pigs. It can also help farmers select animals with reduced risk of becoming lame in the future. High-tech camera systems can provide:

  • a diagnostic tool to quickly identify lame animals
  • a predictive tool to help choose genetically superior animals as future breeding stock

Research project

Using motion capture technology to quantify locomotion in pigs

We started developing the technology with quantification of normal walking patterns in pigs, in biomechanical terms. We placed reflective markers on key anatomical points of the animal, such as on the hoof and shoulder blade, and on specific joints on each leg. The markers were carefully attached to the skin using sticky tape. This allowed them to be easily removed at the end of the session and reused in later weeks to study walking patterns (gait) as the animal grew in size.

Next, we set up a series of up to 14 cameras to track the movement of the pig along a dedicated 'catwalk'. We trained the pigs to walk behind a technician, who was holding a tray containing tempting pieces of apple. Pigs who walked slowly and steadily to the end of the catwalk were rewarded with a piece of apple. The process of training could take several weeks, and was more successful with some pigs than others. But the general rule was true – that pigs will do anything for food!

A standard video camera recorded the pig's movement as a reference, and a series of infra-red cameras recorded the reflection given off by the skin markers. This allowed the pig's movement to be digitally stored. Once back in the office, we were able to extract key biomechanic data.

The process needs to be followed precisely to extract gait parameters, such as changes in the maximum and minimum values for angles on key joints.

Results

Changes in gait due to lameness

Pigs who are lame because of a problem with one of their front legs show a distinctive change in gait. A pig with front leg lameness has a much more pronounced vertical displacement of its head, by as much as 40mm more than a normal pig (Stavrakakis et al, 2015).

In a population of breeding females, we were also able to use gait as a predictor of pigs who later became lame. Step-to-stride length ratio should be 0.5 in normal pigs. It reliably identified up to 74% of pigs which went on to develop lameness later in life (Stavrakakis et al, 2015). Including biomechanical assessment of gait in genetic improvement programmes could lead to a large reduction in lameness in pig herds across the world.

Developing a farm-level gait monitor

We are investigating using the characteristic ‘head bob’ in a lame pig to develop a farm-level version of the technology. Such technology would detect lameness without using reflective skin markers and infra-red cameras.

Depth cameras, such as the Kinect camera developed by Microsoft for its V-Box gaming console, have the potential to track pig gait from above. This time, instead of the catwalk in the gait laboratory, the camera could be positioned above the pigs in a passageway inside a standard pig building. Pigs with a significant head bob would be detected as they pass underneath.

Our pilot study confirms that this simplified system works in theory. We are excited about the prospect of taking this forward into a farm-level version that could interface with other control systems to monitor environmental conditions in the building as well as pig behaviour.